[seqfan] Re: Program syntax and the OEIS

[Charles Greathouse]

> In the backwards-compatibility direction, sometimes users like to stick to
> their earlier versions for good reasons--they don't want to pay upgrade
> costs, they have other irresolvable functionality issues, and so on.

Right -- decisions need to be made on a case-by-case basis.  For
A006506, the code was rewritten: perhaps very few people were using
Maple versions that old.  The GP fix I mentioned is easy -- it's fully
backward-compatible, since && and || have always worked.

My interests here are twofold: to start discussion on how we should
code versions (and to suggest that we should start to do so), and to
find if there's a good way to search the sequences with regular
expressions ('grepping').  This may be computationally expensive, in
which case restricting uses (similar to the superseeker) or even
restricting users, may be appropriate.  Of course in the immediate
term we don't have anything set up to do that either, so unless we set
something like that up soon-ish*, we'd either have to not do it or do
it manually, which will depend on how the sequences are stored on the
server.

* I could help.

Charles Greathouse
Analyst/Programmer
Case Western Reserve University

On Fri, Jun 24, 2011 at 12:10 PM, Marc LeBrun <mlb at well.com> wrote:
>>="Alonso Del Arte" <alonso.delarte at gmail.com>
>> a comment like (* Version 6.0+ *) might be sufficient
>
> Documentation seems the only really viable approach. The OEIS can't be in
> the code-snippet maintenance business.
>
> Automatic machine translation of programs is an interesting research topic,
> but relying on it is dicey in the forward direction.  How can you be *sure*
> that all important semantics are correctly preserved?
>
> In the backwards-compatibility direction, sometimes users like to stick to
> their earlier versions for good reasons--they don't want to pay upgrade
> costs, they have other irresolvable functionality issues, and so on.
>
> It's kind of like respecting folks who want to stick with ASCII-only eMail.
>
> Recoding already-submitted programs for a different version isn't quite the
> same thing as a "correction" (as an actual bugfix would be), and may merely
> shift the burden from one user community to another.
>
> Instead, the instructions for submitting code should at least strongly
> encourage adding a comment giving version info, as a "best practice".
>
> Perhaps this could even be supported somehow as an extension when entering
> the "name" of the CAS, language or other framework?
>
> It's essentially the same as what the "++" in the name "C++" conveys.  It's
> just that some of these systems change more frequently, and choose different
> syntax (there's now also "C++0x").  It sounds to me like Maple5 and Maple6
> are closely related but different languages (as are C and C++).
>
> Instead of recoding the Maple5, why not leave the earlier version in place,
> but supplement it with a new Maple6 version, just as we would add a version
> for C++ or any other additional language?
>
>
>
> _______________________________________________
>
> Seqfan Mailing list - http://list.seqfan.eu/
>